An increasing number of models for predicting land use change in rapidly urbanizing regions are being proposed and built using ideas from cellular automata (CA). Calibrating such models to real situations is highly problematic and to date, serious attention has not been focused on the estimation problem. In this paper, we propose a structure for simulating urban change based on estimating land use transitions using elementary probabilistic methods which draw their inspiration from Bayes’ theory and the related ‘weights of evidence’ approach. These land use change probabilities drive a CA model based on eight cell Moore neighborhoods implemented through empirical land use allocation algorithms. The model framework has been applied to a medium-sized town, Bauru, in the west of São Paulo State, Brazil. We show how various socio-economic and infrastructural factors can be combined using the weights of evidence approach which enables us to predict the probability of changes between land use types in different cells of the system. Different predictions for the town during the period 1979–1988 were generated, and statistical validation was then conducted using a multiple resolution fitting procedure. These modeling experiments support the essential logic of adopting Bayesian empirical methods which synthesize various information about spatial infrastructure as the driver of urban land use change. This indicates the relevance of the approach for generating forecasts of growth for Brazilian cities in particular and for world-wide cities in general.